OpenAI CEO Sam Altman announced in his post on the social media platform X that ChatGPT now complies with user requests not to use the hyphen (—) character. Previously, even though an instruction to this effect was given in the special settings section, ChatGPT ignored this request and continued to frequently use these long dashes, known as em dash, in texts. With this change, it becomes possible to gain more control over ChatGPT’s writing style.
Increasingly, it is emphasized that we need to be more careful about whether online content is created by artificial intelligence or humans. Particularly the heavy use of the hyphen is considered by many as a sign that the text was produced with large language models. Although this writing habit does not constitute evidence on its own, it has become one of the factors taken into consideration in detecting artificial intelligence-supported texts.
ChatGPT updates may not be limited to hyphens
Although the use of hyphens is not unique to artificial intelligence, this punctuation mark appears frequently in texts produced by natural language processing models. This is because large language models known as LLM (Large Language Model); This is due to being trained with large data sets such as books, scientific publications, forum posts and digital articles. This sign, which is widely used in training data, is also automatically included in the model’s language production. Previously, user requests to limit such style preferences were not taken into account by the model. However, this situation seems to have changed with the latest update.
Despite all this, the presence of the hyphen character does not necessarily indicate that a text was written with artificial intelligence. Experienced writers, academic content producers or journalists can also use this sign consciously to distinguish meanings or regulate the flow of sentences. However, the fact that users interpret certain writing patterns in the content as artificial intelligence production necessitates platforms to provide more control over such style preferences.
However, this update by OpenAI may not be limited to just the use of hyphens. The improvement of ChatGPT’s response to user instructions indicates that more comprehensive customization options may be offered in the future. Especially in corporate and academic use, artificial intelligence models that can adapt to writing styles in texts are preferred by a wider range of users.